Gold trading basics · Lesson 56 of 100

Gold Bollinger Bands: Distinguish Price Dispersion from a Reversal Signal

Calculate illustrative bands, bandwidth and percent-B while separating price-level dispersion from return volatility and probability claims.

Bands Measure Dispersion: instructional diagram for lesson 56 — CYGURU GOLD · Original lesson schematic. Examples are illustrative.

Specify the series and the estimator

Bollinger-style bands place an upper and lower boundary around a moving average using a selected multiple of standard deviation. The exact source field, averaging method, window, deviation convention and offset determine the result. A standard deviation of price levels is different from the return-based volatility studied in lesson 45. It describes dispersion of the selected prices around their average, in price units. Record the implementation rather than assuming every platform uses the same denominator or initialization. A familiar indicator name does not settle those choices, particularly when a custom script changes the defaults.

Work through a fully specified toy window

Take four invented gold closes: 1,990, 2,000, 2,010 and 2,000. Their simple mean is 2,000. Squared deviations total 200. For this teaching example, use the population divisor of four, giving variance 50 and standard deviation approximately 7.0711. At a multiplier of two, the upper band is approximately 2,014.1421 and the lower band 1,985.8579. This short window makes the arithmetic inspectable; it is not a recommended production setting. A sample divisor of three would produce different bands. State the divisor explicitly before comparing the worksheet with a chart or claiming that its output is incorrect.

Measure position and width separately

Percent-B can be defined as price minus lower band divided by upper band minus lower band. At price 2,010, the example yields approximately 0.8536. Values above one or below zero are possible when price lies outside the bands. Bandwidth can be defined as upper minus lower divided by the middle average; here it is approximately 1.4142%. These measures answer different questions: location within the band structure and relative width of that structure. Preserve the percent versus fraction convention. A percent-B reading of 0.85 is not an 85% probability of reversal, and bandwidth is not an account-loss estimate.

Band worksheet · population divisor in this example
Measure	Calculation	Result
Mean	(1,990+2,000+2,010+2,000)/4	2,000
Deviation	sqrt(200/4)	7.0711
Upper / lower	2,000 ± 2×sqrt(50)	2,014.1421 / 1,985.8579
Percent-B at 2,010	(price−lower)/(upper−lower)	0.8536
Bandwidth	(upper−lower)/middle	1.4142%

No automatic future-coverage probability follows from these rolling price bands.

Do not turn two deviations into guaranteed coverage

The common interpretation that two standard deviations enclose about 95% of observations comes from particular distributional conditions. Rolling financial price levels are not automatically independent observations from a stationary normal distribution. The bands themselves also change as observations enter and leave the window. A price can remain near an upper band during a sustained advance, and a close outside it does not establish an immediate return to the average. Describe the observed location without importing an unsupported probability. If a strategy assumes mean reversion, that assumption needs separate evidence under its actual entry, exit and cost rules.

Separate a squeeze from a directional forecast

A narrow band structure indicates low dispersion under the selected calculation, compared with an explicitly chosen reference. It does not determine the direction or timing of the next expansion. Defining a squeeze after looking at a dramatic breakout can select a successful example retrospectively. Choose the width threshold, comparison period, trigger and failed-break treatment before evaluation. Check whether a changing mean, a gap or an unusual observation drives the width. A visually appealing squeeze is descriptive information; the research question is whether a fixed rule using it adds value beyond a simpler comparison under realistic execution assumptions.

Practice task: explain what the bands cannot tell you

Recalculate the mean, variance and both boundaries, then derive percent-B and bandwidth from the same unrounded values. Replace 2,010 with 2,030 and observe how the window’s mean and dispersion change together. Do not compare the new price against old bands unless the exercise explicitly freezes them at the previous decision time. Write a short report distinguishing a completed-bar band touch, an intrabar touch and a future reversal. Include the estimator, timing and threshold definition. Your final conclusion should identify the missing trading rule rather than presenting band contact as a complete decision.

Gold Bollinger Bands: Distinguish Price Dispersion from a Reversal Signal

Specify the series and the estimator

Work through a fully specified toy window

Measure position and width separately

Do not turn two deviations into guaranteed coverage

Separate a squeeze from a directional forecast

Practice task: explain what the bands cannot tell you

Can you apply the idea?

Make this lesson part of your process.

Gold Trading Mastery

Continue the beginner course

Lesson sources and further study

Core course references

Connected reading